I'm a Data Scientist and AI/ML Engineer with expertise in Generative AI, Large Language Models, and Cybersecurity. I also serve as a Cyber Operations Officer in the Army National Guard.
With a Master's in AI from Boston University and experience at Capital One, Wells Fargo, and MIT Lincoln Laboratory, I specialize in developing cutting-edge AI solutions that solve real-world problems.
Get in touch AvailableRecently Finished
Multi-generational AI civilization simulator. LLM agents breed, build families, and evolve from empty earth to complex societies. Features real inheritance, environmental dynamics, discovery materials, and 15+ enhanced commands for building artificial civilizations. Learn more.
JudgeFlow is a modular, responsible LLM evaluation framework. Plug in new metrics with YAML, run one command, and get a CSV-grade-sheet & reliability diagram. Features built-in calibration, fairness, privacy, and safety metrics. Learn more.
Open‑sourced a 12 KB Python CLI that adds tamper‑evident signatures, license metadata, and SBOM export to any LoRA / adapter file. Ships with a Hugging Face push filter, CycloneDX generator, and a live docs site. Released on PyPI (v 0.1.2) under Apache‑2.0.
Achieved a top 10% finish (166th out of 1727) and earned a Bronze Medal in Kaggle's highly competitive March Machine Learning Mania 2025. Developed predictive models for NCAA tournament outcomes using a custom ELO system. Read the full write-up for details on my approach!
Currently working on...
Pursuing AWS Certified Solutions Architect certification to deepen expertise in cloud architecture and scalable system design.
Building a complete language model from scratch following Stanford's CS336 course. Implementing data collection and cleaning, transformer architecture, training infrastructure, and evaluation pipelines. Learning every aspect of modern LLM development from tokenization to deployment.
Writing
Follow me for thoughts on AI, machine learning, and tech. I regularly share my thoughts about the latest developments in the field.
Insights from implementing production-grade RAG systems and improving information retrieval performance.
How I ship features at warp speed with Cursor, Gemini 2.5, RepoPrompt & friends.
How I finished 166th out of 1,727 participants using a custom ELO system with strategic manual overrides. Top 5 was within reach!